An empirical study of scoring auctions and quality manipulation corruption

We provide a modeling framework for empirical studies of scoring auction data that addresses the problem that scores are intangible and cannot be compared across auctions. We apply the method on a dataset from Chinese server room procurement auctions and provide empirical evidence for the seminal scoring auction model developed by Che (1993) and Asker and Cantillon (2008). Using this method, we study the problem of quality manipulation: the quality evaluating agent may exaggerate the quality score of a favored firm in exchange for a bribe. We propose a test for quality manipulation and perform it on the server room procurement dataset. We find that the bidding behaviors are generally consistent with the competitive model but not the corruption model.